Hosting · Domain & DNS
The internet's address book, kept current.
Domain registration, transfer, and managed DNS on anycast resolvers. DNSSEC, fast TTLs, and the records-management discipline that prevents the 'who broke MX?' email thread.
DNS is where small mistakes become large outages.
Most of the worst outages we've cleaned up after weren't application failures. They were DNS. Someone updated a record and forgot the matching SPF, so email started bouncing. Someone set a 24-hour TTL and then needed to fail over, so the fix took a day to reach anyone. Someone let the domain quietly expire, and the company vanished from the internet over a weekend. None of these are exotic. They're the ordinary result of treating DNS as a settings page rather than a production system, and the cost is usually a day or two of unreachable email plus the harder-to-measure damage of customers finding you offline.
The fix is unglamorous and durable. We move the DNS onto managed anycast resolvers, document every record and what depends on it, set TTLs deliberately, and put the renewal dates somewhere a person will actually see them before they pass. Then changes go through people who understand the blast radius of each edit. Because DNS is also where your email authentication records live and where your website address resolves, getting this layer right quietly improves the reliability of nearly everything sitting on top of it.
The single best thing they did the first month was move our DNS off the registrar's nameservers. Resolution dropped to single digits.